Skip to content
Snippets Groups Projects
Luca Giommi's avatar
Luca Giommi authored
Fix: add required false for region property in SlaPlacement node (CLOUD-2681)

See merge request !138
3dc6c5bc
History

Intro

TOSCA Templates used by INFN-Cloud PaaS to create instances of supported services.

Kubernetes

Topology template to create a Kubernetes cluster instance.

Input Variables

  • admin_token: Password token for accessing Grafana dashboard
  • number_of_masters: Number of VMs for K8s master
  • num_cpus_master: Number of CPU for K8s master VM
  • mem_size_master: Memory size for K8s master VM
  • number_of_nodes: Number of K8s node VMs
  • num_cpus_node: Number of CPUs for K8s node VM
  • mem_size_node: Memory size for K8s node VM
  • number_of_nodes_with_gpu: Number of K8s node VMs with GPU support
  • num_cpus_node_with_gpu: Number of CPUs for K8s node VM with GPU support
  • mem_size_node_with_gpu: Memory size for K8s node VM with GPU support
  • num_gpus_node: Number of GPUs for K8s node with GPU support
  • gpu_model_node: GPU model
  • enable_gpu: Flag to enable GPU support (apply to GPU accelerated nodes)

Example input

The following command creates a Kubernetes cluster with one node with GPU support and another node without GPU. Notice that you can specify different memory/cpus requirements for nodes with and without GPU.

orchent depcreate -g ${INFN_CLOUD_GROUP} k8s_cluster.yaml \
'{
  "admin_token": "xyz",
  "number_of_masters": 1,
  "mem_size_master": "8 GB",
  "num_cpus_master": 4,
  "number_of_nodes": 1,
  "mem_size_node": "128 GB",
  "num_cpus_node": 16,
  "number_of_nodes_with_gpu": 1,
  "mem_size_node_with_gpu": "64 GB",
  "num_cpus_node_with_gpu": 8,
  "num_gpus_node": 1,
  "gpu_model_node": "T4",
  "enable_gpu": true,
  "users": [{"os_user_add_to_sudoers": true, "os_user_name": "'${INFN_CLOUD_USERNAME}'", "os_user_ssh_public_key": "'${INFN_CLOUD_PUBLIC_KEY}'"}]
}'

Note: if number_of_nodes_with_gpu > 0 and is enable_gpu == false, no software to enable GPU support (e.g. GPU vendor drivers) will be installed on GPU accelerated nodes, this allows users to customize GPU support.

Requirements